Future Research for Wireless Communication Integrated Circuits

1
Future Research for Wireless Communication
Integrated Circuits

• Professor Ali M Niknejad
• Berkeley Wireless Research Center
• University of California, Berkeley

2
Research Team

• Axel Berny1 Wideband VCO
• Mounir Bohsali 60 GHz SiGe Receiver
• Kevin Cao2 Interconnect Analysis
  Inductor Modeling
• Sohrab Emami3 CMOS DeviceModeling
• Hanching Fuh Dual Mode PAs
• Eddie Ng 60 GHz SiGe VCO / dividers
• Shahin Farshchi Dynamic PA
• Vibhav Mittal Extension of ASITIC
• 1 Co-advised with Prof. Meyer
• 2 Co-advised with Prof. Hu and Prof. King
• 3 Co-advised with Prof. Brodersen

3
Show Stoppers in the Wireless Comm Revolution

• Power Consumption
• Multiplicity of Standards
• Requirement for Linear PA
• External Component Count
• Higher Frequency of Operation
• Time to Market

4
Power Consumption

• Mobile applications require very low power
• Cellular transceivers burn 100mW of power
  while power amplifiers burn several watts
• Talk time limited by PA and DSP to a few hours
  (today we can safely ignore transceiver power!)
• In the future the PA power will drop
  significantly (power control will regulate power
  to low levels)
• Transceiver power will thus become equally
  important
• Transceivers today designed for worst-case
  scenario of distant base station in presence of
  nearby interferers

5
Dynamic Transceivers

• High power consumption comes about due to
  simultaneous requirement of high linearity in RF
  front-end and low noise operation
• The conflicting requirements occur since the
  linearity of the RF front-end is exercised by a
  strong interferer while trying to detect a weak
  signal
• The worst case scenario is a rare event. Dont
  be pessimistic!
• A dynamic transceiver can schedule gain/power of
  the front-end for optimal performance

6
Multiplicity of Standards

• Cellular voice GMS, CDMA, W-CDMA, CDMA-2000,
  AMPS, TDMA
• Same Standard over multiple frequency bands (4-5
  GMS bands exist today)
• Data 802.11b, 802.11a, Bluetooth, 3G
• A typical handheld computer or laptop should be
  compatible with all of the above standards

7
Universal Radio

• High Dynamic Range Broadband Front End
• High speed high dynamic range ADC
• Eliminate high-Q front-end filtering
• Design parallel or broadband amplifiers to cover
  major bands around 1 GHz, 2 GHz, 5 GHz, etc.
• Require dynamic operation to reduce power
• Employ broadband matching, filtering, and
  amplification (e.g. 500 MHz 3 GHz)

8
Linear Power Amplifiers

• Todays voice systems employ non-linear PAs due
  to the use of constant-envelope modulation
  schemes
• Spectrally efficient modulation schemes require a
  linear PA
• Wi-Fi requires a linear PA (due to OFDM)
• Linear PAs have relatively poor efficiency lt 30
• Efficiency especially problematic at lower power
  levels (CDMA systems have about 7 average eff.)
• Technology trends in the wrong direction (lower
  supply voltages and lower breakdown)
• PAs typically realized in GaAs or V-MOSFET
  technologies (SiGe, GaN, GaC, others!)

9
Intelligent Power Amplifiers

• PAs today are dumb. They should be more adaptive
• A dynamic matching network can compensate for
  antenna impedance changes due to near-field
  environmental changes and also due to process
  variations on the board and matching components.
• Power levels are dropping for dense urban env
  Need separate PAs optimized for both low power
  and high power must co-exist and interoperate to
  maintain efficiency across entire power range
• Research Efforts
• Dual Mode Class A/F PA project
• Dynamic Class A PA project
• Linearized Class C Amplifier

10
High External Component Count

• Current trends in academia and industry have
  reduced component count at RF and IF
• The Low-IF, Direct-Conversion, and Wideband IF
  radio architectures eliminate (reduce) external
  IF filters
• Systems still heavily dependent on external
  components on the front end SAW filters,
  switches, directional couplers, matching
  networks, diodes, duplexers
• Many of these components are expensive (high Q)
  and narrowband

11
Reducing Front-End Components

• Front-end components are a major impediment to
  the design of a more universal filter
• Integration of more passive elements on-chip or
  in the package
• Broadband front-end with improved linearity to
  cover multiple bands eliminates high-Q filters
• Integrated Matching Networks for PA and LNA
• Need simulation tools to do co-simulation of
  chip, package, and board environment

12
Higher Frequency of Operation

• The typical Internet user is very bandwidth
  hungry real-time high fidelity digital music
  requires 100 kbps/s. Video will require
  several Mbps/s.
• New high frequency bands offer a healthy supply
  of bandwidth (5 GHz around 60 GHz)
• Current CMOS RF-ICs and design methodology is
  ill-equipped to move into these bands
• Traditional Microwave methodology also
  inappropriate (You want to use more than ten
  transistors?)
• Current EM Solvers are not capable of handling
  VLSI circuits (LSI?)

13
60 GHz CMOS WLAN

• Large team attacking problem of designing full
  communication system antennas, filters,
  amplifiers, mixers, and baseband circuitry.
• Two paths pure CMOS and SiGe BiCMOS
• Microwave circuit techniques essential but not
  sufficient
• New circuit design techniques requires for near
  fT operation
• CMOS models to account for distributed effects

14
Wideband VCO

• Broadband systems require VCO to tune over a
  significant range (say 100 of carrier)
• Varactors only provide 20 tuning range
• Switched capacitor techniques extend tuning range
  (coarse tuning fine tuning)
• Employing multiple resonant modes can also
  increase tuning range
• Special techniques necessary to set oscillation
  amplitude over wide frequency range (cant rely
  solely on amplitude limiting)

15
RF-ICs Time to Market

• RF expertise hard to find
• Typical IC manufacturer spend 2 years in
  development, testing, and manufacture of an
  advanced RFIC
• The consumer market is hungry for products today!
  (Wheres my wireless USB port on my laptop?)
• RFICs are supposed to be cheap!
• CAD tools and a lack of expertise is a major
  detriment in fast implementation of todays
  systems

16
Transceiver Optimization

• Typical transceivers designed by large groups of
  engineers
• Typically the transceiver is divided among
  several blocks LNA, VCO, PA, mixer, etc.
• Optimizing individual blocks separately does not
  yield global optimum!
• Co-design of tightly coupled blocks
  (LNAmixerbuffer)
• Dynamic power allocation requires a holistic
  viewpoint, not an atomistic approach to the
  optimization

17
Microprocessors are Becoming Microwave Circuits!

• Low jitter GHz clock lines difficult to implement
  and consume a lot of power
• Due to large capacitance of loads and line, must
  drive C V 2 f power
• Take advantage of distributed nature of clock
  line inductance of line and capacitance of
  loads form artificial transmission line
• Power drive to clock resonant network reduced
  significantly if network has high Q
• Reduce substrate parasitics, eddy currents, etc.